Simulation results - overall

These simulations used prior specifications that involved a large prior SS, i.e. \(\sigma^2 \sim IG(1,100)\). While this prior is quite large given the true variance in the data-generating process, \(\sigma^2 = 10\), it appears in the well-separated case to result in improved clustering performance under the DPMM.

## `summarise()` has grouped output by 'Model', 'Scenario', 'SM'. You can override
## using the `.groups` argument.
Adjusted RAND Index
Model SM 3wellsep_30 3wellsep_100 3wellsep_300
conjDEE noSM 0.95 (0.93,0.97) 0.8 (0.71,0.85) 0.69 (0.56,0.88)
conjDEE withSM 0.94 (0.92,0.96) 0.94 (0.88,0.96) 0.85 (0.69,0.98)
conjDEV noSM 0.99 (0.98,1) 1 (0.86,1) 1 (0.86,1)
conjDEV withSM 0.98 (0.96,0.99) 0.99 (0.97,0.99) 0.99 (0.98,1)
## `summarise()` has grouped output by 'Model', 'Scenario', 'SM'. You can override
## using the `.groups` argument.
KL Divergence
Model SM 3wellsep_30 3wellsep_100 3wellsep_300
conjDEE noSM 0.07 (0.05,0.09) 0.03 (0.02,0.04) 0.01 (0.01,0.02)
conjDEE withSM 0.07 (0.05,0.09) 0.02 (0.01,0.03) 0.01 (0,0.01)
conjDEV noSM 0.08 (0.06,0.09) 0.03 (0.02,0.04) 0.01 (0.01,0.01)
conjDEV withSM 0.07 (0.06,0.09) 0.02 (0.02,0.03) 0.01 (0,0.01)
## `summarise()` has grouped output by 'Model', 'Scenario', 'SM'. You can override
## using the `.groups` argument.
MAP(K)
Model SM 3wellsep_30 3wellsep_100 3wellsep_300
conjDEE noSM 3 (3,3) 4 (4,5) 5 (3,7)
conjDEE withSM 3 (3,3) 3 (3,3) 3 (3,3)
conjDEV noSM 3 (3,3) 3 (3,4) 3 (3,4)
conjDEV withSM 3 (3,3) 3 (3,3) 3 (3,3)

Selected traceplots - without split/merge

n = 30

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]   -1
## [2,]    1
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 3.097688
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] FALSE
## 
## $sm_iter
## [1] 0

n = 100

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]    0
## [2,]    0
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 18.20804
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] FALSE
## 
## $sm_iter
## [1] 0

n = 300

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]    0
## [2,]    0
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 3.104484
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] FALSE
## 
## $sm_iter
## [1] 0

Selected traceplots - with split/merge

n = 30

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]   -1
## [2,]    1
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 7.94319
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] TRUE
## 
## $sm_iter
## [1] 10

n = 100

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]    0
## [2,]    0
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 2.646533
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] TRUE
## 
## $sm_iter
## [1] 10

n = 300

## $S
## [1] 12000
## 
## $alpha
## [1] 1
## 
## $a
## [1] 1
## 
## $b
## [1] 100
## 
## $mu0
##      [,1]
## [1,]    0
## [2,]    0
## 
## $k_init
## [1] 1
## 
## $init_method
## [1] "kmeans"
## 
## $d
## [1] 1
## 
## $f
## [1] 1
## 
## $g
## [1] 1
## 
## $h
## [1] 25
## 
## $r
## [1] 5.259726
## 
## $mod_type
## [1] "conjDEE"
## 
## $split_merge
## [1] TRUE
## 
## $sm_iter
## [1] 10